Skip to content

HDDS-13148. [Docs] Update Transparent Data Encryption doc. #8530

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
smengcl merged 8 commits into from
Jun 11, 2025
Merged

HDDS-13148. [Docs] Update Transparent Data Encryption doc. #8530

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
2bd55a6
HDDS-13148. [Docs] Update Transparent Data Encryption doc.
jojochuang May 30, 2025
3dcd627
Update
jojochuang May 30, 2025
834ef09
Update per Gemini.
jojochuang Jun 4, 2025
b6bc2c6
Update per Siyao's comments.
jojochuang Jun 4, 2025
60fe96a
Update hadoop-hdds/docs/content/security/SecuringTDE.md
jojochuang Jun 4, 2025
3633e73
Update per Gemini
jojochuang Jun 4, 2025
dd1fca0
Update hadoop-hdds/docs/content/security/SecuringTDE.md
jojochuang Jun 4, 2025
f17d50e
Revert unrelated change.
jojochuang Jun 4, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
150 changes: 92 additions & 58 deletions hadoop-hdds/docs/content/security/SecuringTDE.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,64 +25,81 @@ icon: lock
limitations under the License.
-->

Ozone TDE setup process and usage are very similar to HDFS TDE.
The major difference is that Ozone TDE is enabled at Ozone bucket level
when a bucket is created.
Ozone Transparent Data Encryption (TDE) enables you to encrypt data at rest. TDE is enabled at the bucket level when a bucket is created. To use TDE, an administrator must first configure a Key Management Server (KMS). Ozone can work with **Hadoop KMS** and **Ranger KMS**. The KMS URI needs to be provided to Ozone via the `core-site.xml` configuration file.

### Setting up the Key Management Server
Once the KMS is configured, users can create an encryption key and then create an encrypted bucket using that key. All data written to an encrypted bucket will be transparently encrypted on the server-side, and data read from the bucket will be transparently decrypted.

To use TDE, admin must setup a Key Management Server and provide that URI to
Ozone/HDFS. Since Ozone and HDFS can use the same Key Management Server, this
configuration can be provided via *core-site.xml*.
### Configuring TDE

Property| Value
-----------------------------------|-----------------------------------------
hadoop.security.key.provider.path | KMS uri. <br> e.g. kms://http@kms-host:9600/kms
1. **Set up a Key Management Server (KMS):**
* **Hadoop KMS:** Follow the instructions in the [Hadoop KMS documentation](https://hadoop.apache.org/docs/r3.4.1/hadoop-kms/index.html).
* **Ranger KMS:** Ranger KMS can also be used. For Ranger KMS, encryption keys can be managed via the Ranger KMS management console or its [REST API](https://ranger.apache.org/kms/apidocs/index.html), in addition to the `hadoop key` command line interface.

### Using Transparent Data Encryption
If this is already configured for your cluster, then you can simply proceed
to create the encryption key and enable encrypted buckets.
2. **Configure Ozone:**
Add the following property to Ozone’s `core-site.xml`:

To create an encrypted bucket, client need to:
<property>
<name>hadoop.security.key.provider.path</name>
<value>kms://http@kms-host:9600/kms</value>
</property>

* Create a bucket encryption key with hadoop key CLI, which is similar to
how you would use HDFS encryption zones.
Replace `kms://http@kms-host:9600/kms` with the actual URI of your KMS. For example, `kms://[email protected]:9600/kms`

```bash
hadoop key create enckey
```
The above command creates an encryption key for the bucket you want to protect.
Once the key is created, you can tell Ozone to use that key when you are
reading and writing data into a bucket.
### Creating an Encryption Key

* Assign the encryption key to a bucket.
Use the `hadoop key create` command to create an encryption key in the configured KMS:

```bash
ozone sh bucket create -k enckey /vol/encryptedbucket
```
```shell
hadoop key create <key_name> [-size <key_bit_length>] [-cipher <cipher_suite>] [-description <description>]
```

* `<key_name>`: The name of the encryption key.
* **`-size <key_bit_length>` (Optional):** Specifies the key bit length. The default is 128 bits (defined by `hadoop.security.key.default.bitlength`).
Ranger KMS supports both 128 and 256 bits. Hadoop KMS is also commonly used with 128 and 256 bit keys; for specific version capabilities, consult the Hadoop KMS documentation. Valid AES key lengths are 128, 192, and 256 bits.
* **`-cipher <cipher_suite>` (Optional):** Specifies the cipher suite. Currently, only **`AES/CTR/NoPadding`** (the default) is supported.
* `-description <description>` (Optional): A description for the key.

For example:

```shell
hadoop key create enckey -size 256 -cipher AES/CTR/NoPadding -description "Encryption key for my_bucket"
```

### Creating an Encrypted Bucket

After this command, all data written to the _encryptedbucket_ will be encrypted
via the enckey and while reading the clients will talk to Key Management
Server and read the key and decrypt it. In other words, the data stored
inside Ozone is always encrypted. The fact that data is encrypted at rest
will be completely transparent to the clients and end users.
Use the Ozone shell `ozone sh bucket create` command with the `-k` (or `--key`) option to specify the encryption key:

```shell
ozone sh bucket create --key <key_name> /<volume_name>/<bucket_name>
```

For example:

```shell
ozone sh bucket create --key enckey /vol1/encrypted_bucket
```

Now, all data written to `/vol1/encrypted_bucket` will be encrypted at rest. As long as the client is configured correctly to use the key, such encryption is completely transparent to the end users.

### Using Transparent Data Encryption from S3G

There are two ways to create an encrypted bucket that can be accessed via S3 Gateway.
Ozone’s S3 Gateway (S3G) allows you to access encrypted buckets. However, it's important to note that **Ozone does not support S3-SSE (Server-Side Encryption) or S3-CSE (Client-Side Encryption) in the way AWS S3 does.** That said, Ozone S3 buckets can be encrypted using Ranger KMS or Hadoop KMS to provide a guarantee similar to S3-SSE with client-supplied keys (SSE-C).

#### Option 1. Create a bucket using shell under "/s3v" volume
When creating an encrypted bucket that will be accessed via S3G:

```bash
ozone sh bucket create -k enckey --layout=OBJECT_STORE /s3v/encryptedbucket
```
1. **Create the bucket under the `/s3v` volume:**
The `/s3v` volume is the default volume for S3 buckets.

#### Option 2. Create a link to an encrypted bucket under "/s3v" volume
```shell
ozone sh bucket create --key <key_name> /s3v/<bucket_name> --layout=OBJECT_STORE
```

```bash
ozone sh bucket create -k enckey --layout=OBJECT_STORE /vol/encryptedbucket
ozone sh bucket link /vol/encryptedbucket /s3v/linkencryptedbucket
```
2. **Alternatively, create an encrypted bucket elsewhere and link it:**

```shell
ozone sh bucket create --key <key_name> /<volume_name>/<bucket_name> --layout=OBJECT_STORE
ozone sh bucket link /<volume_name>/<bucket_name> /s3v/<link_name>
```

Note 1: An encrypted bucket cannot be created via S3 APIs. It must be done using Ozone shell commands as shown above.
After creating an encrypted bucket, all the keys added to this bucket using s3g will be encrypted.
Expand All @@ -94,12 +111,12 @@ argument, but explicitly added here to make a point).
Bucket created with the `OBJECT_STORE` type will NOT be accessible via
HCFS (ofs or o3fs) at all. And such access will be rejected. For instance:

```bash
```bash
$ ozone fs -ls ofs://ozone1/s3v/encryptedbucket/
-ls: Bucket: encryptedbucket has layout: OBJECT_STORE, which does not support file system semantics. Bucket Layout must be FILE_SYSTEM_OPTIMIZED or LEGACY.
```

```bash
```bash
$ ozone fs -ls o3fs://encryptedbucket.s3v.ozone1/
22/02/07 00:00:00 WARN fs.FileSystem: Failed to initialize fileystem o3fs://encryptedbucket.s3v.ozone1/: java.lang.IllegalArgumentException: Bucket: encryptedbucket has layout: OBJECT_STORE, which does not support file system semantics. Bucket Layout must be FILE_SYSTEM_OPTIMIZED or LEGACY.
-ls: Bucket: encryptedbucket has layout: OBJECT_STORE, which does not support file system semantics. Bucket Layout must be FILE_SYSTEM_OPTIMIZED or LEGACY.
Expand All @@ -112,37 +129,54 @@ However, in buckets with `FILE_SYSTEM_OPTIMIZED` layout, some irregular S3 key
names may be rejected or normalized, which can be undesired.
See [Prefix based File System Optimization]({{< relref "../feature/PrefixFSO.md" >}}) for more information.

In non-secure mode, the user running the S3Gateway daemon process is the proxy user,
while in secure mode the S3Gateway Kerberos principal (ozone.s3g.kerberos.principal) is the proxy user.
S3Gateway proxy's all the users accessing the encrypted buckets to decrypt the key.
For this purpose on security enabled cluster, during S3Gateway server startup
logins using configured
**ozone.s3g.kerberos.keytab.file** and **ozone.s3g.kerberos.principal**.
When accessing an S3G-enabled encrypted bucket:

The below two configurations must be added to the kms-site.xml to allow the S3Gateway principal to act as a proxy for other users. In this example, "ozone.s3g.kerberos.principal" is assumed to be "s3g"
The below three configurations must be added to the kms-site.xml to allow the S3Gateway principal to act as a proxy for other users. In this example,
`ozone.s3g.kerberos.principal` is assumed to be `s3g`

```xml
<property>
<name>hadoop.kms.proxyuser.s3g.users</name>
<value>user1,user2,user3</value>
<description>
Here the value can be all the S3G accesskey ids accessing Ozone S3
or set to '*' to allow all the accesskey ids.
Specifies the list of users that the S3 Gateway (`s3g`) is allowed to impersonate when interacting with the KMS. Use `*` to allow all users.
</description>
</property>
<property>
<name>hadoop.kms.proxyuser.s3g.groups</name>
<value>group1,group2,group3</value>
<description>
Specifies the list of groups whose members `s3g` is allowed to impersonate when making requests to the KMS. Use `*` to allow all groups.
</description>
</property>

<property>
<name>hadoop.kms.proxyuser.s3g.hosts</name>
<value>s3g-host1.com</value>
<description>
This is the host where the S3Gateway is running. Set this to '*' to allow
requests from any hosts to be proxied.
Specifies the hostnames or IPs from which `s3g` is permitted to send proxy requests to the KMS. Use `*` to allow all hosts.
</description>
</property>
```

### KMS Authorization

If Ranger authorization is enabled for KMS, then decrypt key permission should be given to
access key id user(currently access key is kerberos principal) to decrypt the encrypted key
to read/write a key in the encrypted bucket.
Key Management Servers (KMS) may enforce key access authorization. **Hadoop KMS supports ACLs (Access Control Lists) for fine-grained permission control, while Ranger KMS supports Ranger policies for encryption keys.** Ensure that the appropriate users have the necessary permissions based on the KMS type in use.

For example, when using Ranger KMS for authorization, to allow the user `om` (the Ozone Manager user) to access the key `enckey` and the user `hdfs` (a typical HDFS service user) to manage keys, you might have policies in Ranger KMS like:

* **Policy for `om` user (or the user running the Ozone Manager):**
* Resource: `keyname=enckey`
* Permissions: `DECRYPT_EEK` (Decrypt Encrypted Encryption Key)
* **Policy for S3 Gateway proxy user (e.g., the user specified in `ozone.s3g.kerberos.principal`, typically `s3g`):**
* Resource: `keyname=enckey` (or specific keys for S3 buckets)
* Permissions: `DECRYPT_EEK`
* **Policy for administrative users (e.g., `hdfs` or a keyadmin group):**
* Resource: `keyname=*` (or specific keys)
* Permissions: `CREATE_KEY`, `DELETE_KEY`, `GET_KEYS`, `ROLL_NEW_VERSION`

Refer to the Ranger documentation for detailed instructions on configuring KMS policies if you are using Ranger KMS. For Hadoop KMS, consult its [Hadoop KMS documentation](https://hadoop.apache.org/docs/r3.4.1/hadoop-kms/index.html#ACLs_.28Access_Control_Lists.29) for managing ACLs.

### Additional References

* For more background on Transparent Data Encryption concepts, you can refer to the [Transparent Encryption in HDFS documentation](https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/TransparentEncryption.html).
* For detailed information on Hadoop KMS, see the [Hadoop KMS documentation](https://hadoop.apache.org/docs/r3.4.1/hadoop-kms/index.html).